Recategorizing Interdisciplinary Articles using Natural Language Processing and Machine/Deep Learning

Interdisciplinarity has become popular due to the novel approaches in academic research and papers as well as solutions to various social problems. Thus, research into interdisciplinary research itself and into interdisciplinarity has also become popular. However, it is difficult to identify a research discipline precisely, which is necessary for the evaluation of ‘interdisciplinarity.’ For example, in general, research disciplines are categorized in accordance with the journals that cater to them. Thus, an important multidisciplinary science journal, such as Nature or Science, does not belong to the category of "general research disciplines." Hence, a new category of "multidisciplinary science" is required for the fundamental evaluation of interdisciplinarity." In this research, a new method of classification is conducted by analyzing the abstracts of research papers using machine learning and deep learning. All the data used is gathered from Web of Science (PY: 2016). The new classifier identifies research disciplines including "multidisciplinary science," in contrast to other current methodology.